Thesis breaks
29%If the story breaks: no measured cashflow to catch it, survival scores 5/10. Re-rates toward the floor (-56%).
): Real Talk valuation
One of the few DePIN compute networks with genuinely real paying demand, and still a tiny business under a mid-cap valuation. Akash is a marketplace for renting GPU and compute (a permissionless, cheaper alternative to AWS/GCP), tenants actually pay for leases on-chain, and the March-2026 Burn-Mint Equilibrium upgrade now buys and burns AKT in proportion to compute spend, a real usage→scarcity link. It pairs thematically with Bittensor as a decentralised-AI play. But the honest numbers are small and choppy. ~$3.1M of GROSS lease spend in FY2025 (the protocol's net take is a fraction of that), against a low-hundreds-of-millions cap, which is a steep multiple, and lease revenue actually FELL sharply after the BME upgrade and trended down through late 2025. The "deflationary" framing is conditional, because at current spend the token still net-inflates and only high adoption flips it. Provider and GPU counts are small (~60-70 providers), and it's competing with AWS, GCP and CoreWeave who have vastly more capital and reliability. Real product, real demand, real token sink, just not yet at a scale that supports the price.
Each run picks a scenario by its odds, then jitters the assumptions (lognormal). The result is a probability distribution, not a price target. Twist the dials.
↓ Twist the dials in the bar pinned at the bottom. The histogram, the cone and the payoff ladder all move as you scroll.
These are "what-if" stories, not forecasts. Each line asks: if adoption played out a certain way, what might the journey look like? Price drifts while adoption is just a promise, steps up if/when the catalyst actually lands, then settles. Dark band = the likely range (middle 50% of modelled outcomes); faint band = the wild 5–95% tail. Every path is one hypothetical of many, driven entirely by the dials and our assumptions, never a prediction or a price target.
These 7 scores are our published read. They're what drive the scenarios above (this is a fixed assessment, not a slider). "Good bet" ≠ "good project": a weak project at a tiny price can still be an asymmetric bet, and the ladder shows how thin the moonshot really is.
Explicit, arguable assumptions. Probabilities are weighted to be real: the modal outcome is sideways, the upside is a tail.
If the story breaks: no measured cashflow to catch it, survival scores 5/10. Re-rates toward the floor (-56%).
The honest middle: the price leans on narrative more than fundamentals (fundamentals 4.9/10 vs narrative 5/10). Lands +6%.
Delivers a good chunk of the promise — re-rates partway to peer parity (+58%). Needs the delivery (5/10) to actually show up.
Delivers everything → re-rates toward what a delivering peer is worth (+135%). Thin odds, gated by a 5/10 delivery score — a call option, not a base case.
Everything in Full peer parity (full delivery) — but in a peak $10T total market instead of today’s ~$2.6T. Same coin, bigger pie: it holds ~0.02% of the market. The other four cards all assume today’s market size; this is the only one that lets the whole tide come in.
The locked % and swing chips are fixed assumptions - identical across all four scenarios.
Akash Network earns roughly $3.0M/yr in real, measurable network revenue. At today's $191.0M cap you're paying 64× sales (a rich multiple already) - the rare coin where cashflow genuinely underpins the price. Here's the rest of what's baked in:
Previous ATH: $8.08 (~$1.60B cap, ×8.4 from today) - ~$1.5-1.7B at the 2021 peak, which predates the AI/DePIN narrative (~294M circ now). Down ~90% from ATH.
Real peers doing the same thing - the ladder the price is betting on, not a forecast.
Bottom line: IF BME burn outpaces emissions as GPU/AI demand grows, Akash re-rates toward Render's brand-tier (low $-billions). Its own ~$1.6B ATH cap is the moon anchor. It has genuine measured revenue (~$3M), more than most DePIN peers, but the post-BME revenue drop is a real warning. Delivering-peer ceiling sits ×8.4 above today - and that needs everything to go right.
Scores read TODAY; these two skate to where the puck is heading - and they (not the scores) move the distribution.
Trajectory 0 flatHonest flat: BME buy-and-burn live, AkashML shipping, enterprise networking launched 30-May-2026 - but lease revenue fell ~45% post-BME, providers down to a recent low, GPU usage -57%. Burn working against a shrinking base.
Community heat 5/10+0.9% favourable lean applied to the fundamentals (survival-gated, capped at 5%) - a nod to the crowd, not a thumb on the price.
What the bulls say: "BME is live so every dollar of compute spend now buys and burns AKT - with AkashML undercutting hyperscalers and enterprise VPC launching, real GPU demand finally drives token scarcity."
Our read: Partly - BME is genuinely live and AKT has real measured revenue (~$3M/yr). But the bull case ignores that lease revenue fell ~45% and usage dropped: conditional scarcity, not delivered scarcity.
Stewardship 7/10sound stewardship - the unproven upside gets the benefit of the doubt.
Lead: Greg Osuri (doxxed founder/CEO, Overclock Labs) + co-founders. Mainnet since Sep-2020; testified to US Congress on decentralised compute (2025).
Track record: Genuinely good - made the CPU-to-GPU AI pivot early, launched AkashML (Nov-2025) at 70-85% below hosted inference. Multi-year shipping, no abandonment.
Alignment: Reasonable - real measured revenue (~$3.1M FY2025), BME (Mar-2026) routes spend to buy-and-burn. Inflationary via staking; BME net-deflationary only at high spend.
Red flags: Lease revenue fell ~45% post-BME (demand fragility, not misconduct); small provider base. No regulatory/conduct red flags.
Anchors: CoinGecko, as of 2026-06-04. Model: open assumptions in src/data/tokens.ts. Built by Elle.
0x0FA8...E4E1 Any EVM chain · ETH · BNB · Polygon · Base The author may hold positions in coins covered here. Tips appreciated, never expected.
Any EVM chain — ETH, BNB, Polygon, Base & more
0x0FA8...E4E1 The author may hold positions in coins covered on this site. Tips are appreciated, never expected.